Scrape HAProxy's status page with ruby and nokogiri

haproxy-scraper.rb

require 'open-uri'
require 'nokogiri'

URL = ARGV.shift
doc = Nokogiri::HTML(open(URL))

tables   = doc.css(".tbl")
services = Hash.new

# every even table is a header
# every odd table is the status body
tables.each_with_index do |header_table,i|
  next if i.odd?

  status = {}
  services[header_table.css('.pxname').children.to_s] = status

  # new header per header_table
  status_header = []

  # body
  header_table.next_sibling.css('tr').each_with_index do |tr,j|

    # gather header
    if tr['class'].match(/titr/)
      if j == 1
        status_header = tr.css('th').collect { |th| th.content.downcase }
      end
    else
      server_status = {}

      tr.css('td').each_with_index do |td, k|

        # remove spans
        td.children.css('span').each { |span| span.replace span.content }
        text = td.content

        if k == 0
          status[text] = server_status
        else

          # -1 because server name has no th
          server_status[status_header[k-1]] = text
        end
      end
    end
  end
end

puts services.inspect